--- name: ijfw-cross-audit description: "Generate a cross-platform multi-model audit (Trident) on a diff, brief, or artifact. Trigger: 'cross audit', 'Trident', 'second opinion', 'check with other models', 'check with other AIs', 'cross-check this', 'get another perspective', /cross-audit" --- ## Execution 1. **Detect artifact.** Accept: diff, file path, brief text, or `HEAD~1..HEAD`. If none provided, ask once: `What should I audit? (diff, file, or paste text)` 2. **Detect auditors.** Check PATH for `codex` and `gemini`. Default: one OpenAI-family + one Google-family, excluding caller's family. Cap at 4 auditors total. 3. **Dispatch in parallel via background bash** -- never hand off to user. Prompt each: security findings, logic issues, reliability concerns, test gaps. ```bash codex "Review for security, logic, reliability, test gaps: " & gemini "Review for security, logic, reliability, test gaps: " & wait ``` 4. **Reconcile.** Deduplicate. Classify: - CONSENSUS -- flagged by 2+ auditors - CONTESTED -- flagged by 1 auditor only - PASS -- no issues 5. **Emit report** using the format below. ## Report format rule (all cross-audit outputs) Any reconciliation report presenting multiple paths MUST use this structure: ``` VERDICT OPTIONS A -- : B -- : REVIEWER CONVERGENCE : "" => Option -- RECOMMENDATION Option -- because . NEXT ACTION ``` Never write "Option A" or "Option B" without "-- name" immediately after it on the same line. ## Output contract Emit a `gate-result` block as the **LAST** content of your output. Nothing after it. Use `gate="cross-audit"`. Statuses: `PASS | CONDITIONAL | WARN | FLAG | FAIL`. Format: ```gate-result { "schema_version": "1.0", "gate": "cross-audit", "status": "", "project_type": "", "lenses": [], "affected_artifacts": [], "accounting": {"duration_ms": 0, "lenses_invoked": 0, "cost_usd": null}, "remediation": [], "receipts_ref": null, "supersedes": null, "gate_id": "--", "emitted_at": "" } ```